Synthesis of core/shell nanocrystals with ordered intermetallic single-atom alloy layers for nitrate electroreduction to ammonia

Structurally ordered intermetallic nanocrystals (NCs) and single-atom catalysts (SACs) are two emerging catalytic motifs for sustainable chemical production and energy conversion. However, both have synthetic limitations which can lead to the aggregation of NCs or metal atoms. Single-atom alloys (SAAs), which contain isolated metal atoms in a host metal, can overcome the aggregation concern because of the thermodynamic stabilization of single atoms on host metal surfaces. Here we report a direct solution-phase synthesis of Cu/CuAu core/shell NCs with tunable SAA layers. This synthesis can be extended to other Cu/CuM (M = Pt, Pd) systems, in which M atoms are isolated in the copper host. Using this method, the density of SAAs on a copper surface can be controlled, resulting in both low and high densities of single atoms. Alloying gold into the copper matrix introduced ligand effects that optimized the chemisorption of *NO3 and *N. As a result, the densely packed Cu/CuAu material demonstrated a high selectivity toward NH3 from the electrocatalytic nitrate reduction reaction with an 85.5% Faradaic efficiency while maintaining a high yield rate of 8.47 mol h−1 g−1. This work advances the design of atomically precise catalytic sites by creating core/shell NCs with SAA atomic layers, opening an avenue for broad catalytic applications. Well-defined single-atom alloy (SAA) nanocrystals possess isolated atom centres and tunable electronic properties but are challenging to synthesize. Here, a direct solution-phase synthesis of Cu/CuAu core/shell nanocubes with tunable SAA layers is reported. The Cu/CuAu nanomaterial is highly active for the electrocatalytic conversion of nitrate into ammonia.

Structurally ordered intermetallic nanocrystals (NCs) and single-atom catalysts (SACs) are two emerging catalytic motifs for sustainable chemical production and energy conversion. However, both have synthetic limitations which can lead to the aggregation of NCs or metal atoms. Singleatom alloys (SAAs), which contain isolated metal atoms in a host metal, can overcome the aggregation concern because of the thermodynamic stabilization of single atoms on host metal surfaces. Here we report a direct solution-phase synthesis of Cu/CuAu core/shell NCs with tunable SAA layers. This synthesis can be extended to other Cu/CuM (M = Pt, Pd) systems, in which M atoms are isolated in the copper host. Using this method, the density of SAAs on a copper surface can be controlled, resulting in both low and high densities of single atoms. Alloying gold into the copper matrix introduced ligand effects that optimized the chemisorption of *NO 3 and *N. As a result, the densely packed Cu/CuAu material demonstrated a high selectivity toward NH 3 from the electrocatalytic nitrate reduction reaction with an 85.5% Faradaic efficiency while maintaining a high yield rate of 8.47 mol h −1 g −1 . This work advances the design of atomically precise catalytic sites by creating core/shell NCs with SAA atomic layers, opening an avenue for broad catalytic applications.
Access to nanoscale multifunctionality and materials with synergistic properties requires the development of heterostructures that assemble nanomaterials with distinctive natures. As an example of heterostructured nanomaterials, well-defined core/shell metal nanocrystals (NCs) create interfaces between chemically and structurally dissimilar materials and demonstrate tailorable, synergistic functionality from a spatially controlled distribution of chemical compositions [1][2][3] . Those core/shell NCs often exhibit enhanced or even unconventional physicochemical properties and thus provide new opportunities for many energy-related catalytic processes, such as the reactions involved in fuel Article https://doi.org/10.1038/s44160-023-00258-x for the electrocatalytic nitrate reduction reaction (NO 3 RR). The NO 3 RR offers a promising NH 3 production route alternative to N 2 reduction because it utilizes NO 3 − pollutants as the nitrogen source, circumventing the activation of the strong N≡N triple bond 29,30 . When coupled with renewable electricity, the NO 3 RR represents an environmentally friendly and energy-efficient route for the recirculation of nitrogen species into the nitrogen cycle and the nitrogen economy 31,32 . Our Cu/ CuAu ordered SAA catalysts demonstrate high selectivity toward NH 3 from the NO 3 RR with a Faradaic efficiency of 85.5% at −0.5 V versus the reversible hydrogen electrode (RHE) and an exceedingly high yield rate of 8.47 mol h −1 g −1 at −0.6 V versus RHE. Furthermore, Cu/CuAu ordered SAA exhibits catalytic stability for 20 consecutive electrolysis cycles. Density functional theory (DFT) calculations, strain analysis from high-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM) and operando differential electrochemical mass spectrometry (DEMS) suggest that the high activity of the Cu/CuAu catalysts can be attributed to the {001}-oriented copper site ensemble strengthening the *NO 3 binding due to an upshift in the d centre of surface copper atoms while weakening the *N anchoring due to strong repulsive interactions from a subsurface gold ligand or surface gold single atom.

Synthesis and characterization of Cu/CuAu SAA nanocubes
The synthesis of precisely controlled deposition of gold single atoms and ordered intermetallic CuAu on copper nanocubes (denoted as Cu/ CuAu SAA and Cu/CuAu ordered SAA, respectively) was carried out using a facile seed-mediated method, as illustrated schematically in Fig. 1. First, uniform copper nanocubes with an edge length of 40 ± 3 nm were synthesized by a modified method reported previously (Supplementary Fig. 1) 33 . A controlled amount of HAuCl 4 in oleylamine was then dropwise injected into the solution and incubated for 30 min (Methods). The galvanic replacement between the metallic copper and Au 3+ induced the Kirkendall effect, promoting atom interdiffusion and rearrangement. The success of our synthesis relies on the dropwise injection of the gold precursor at elevated temperature, which allows gold atoms to instantaneously spread across the copper surface, resulting in the formation of well-dispersed gold single atoms on copper nanocubes. The amount of gold precursor injected plays a crucial role in determining the density of gold single atoms, shell thickness and the structure of gold SAAs on copper. The morphology and structure of the particles were then confirmed by bright-field (BF) transmission electron microscopy (TEM) and HAADF-STEM imaging, as shown in Fig. 2. The as-synthesized Cu/CuAu SAA with ~3.7 wt% gold contains nanocubes with a uniform size of 44 ± 4 nm (Fig. 2a). Zoomed views of an individual Cu/CuAu SAA particle (Fig. 2b,c) show atomic planes with cells (oxygen reduction reaction and fuel oxidation reactions) 4-6 , watersplitting cells (hydrogen evolution and oxygen evolution reactions) 7,8 , and small molecule transformation schemes (CO 2 reduction) 9,10 . Ideally, core/shell metal NCs need to be fabricated with a low-cost metal core with precious metal atoms in a thin (≤1 nm) shell to enhance atom efficiency and to tune the properties of NCs through interfacial electronic and geometric effects 11,12 . Tailoring the thin shell in an ordered intermetallic structure with long-range atomic ordering and strong d-d orbital coupling could open up new avenues for improving the catalytic properties of core/shell NCs [13][14][15] . However, high temperatures are involved in the phase transformation, leading to synthetic difficulty in creating such a well-defined, atomic precise structure.
The development of single-atom catalysts (SACs) has emerged as an effective strategy to maximize the atom efficiency of precious metals, and has seen a recent surge of interest 16,17 . In contrast to conventional SACs, which suffer from a lack of control over single-atom loading and aggregation due to the Gibbs-Thomson effect 18,19 , single-atom alloys (SAAs) with atomically dispersed metal atoms in a host metal thermodynamically stabilize single atoms while still embracing isolated atom centres and tunable electronic properties for enhanced catalysis [20][21][22] . Moreover, ordered intermetallic structures can be leveraged to break the continuous metal ensembles, representing a structural motif with a high density of isolated atoms because the first nearest neighbour of a single atom is the host metal 23,24 . For example, in the intermetallic Pm3 m PdIn, palladium atoms are completely isolated by indium atoms 24 . Intermetallic C2/m Al 13 Fe 4 contains iron atoms entirely isolated by aluminium 25 . Lead-modified intermetallic P2 1 3 PtGa with single platinum atoms isolated by gallium at the surface exhibits high stability and selectivity for propane dehydrogenation 26 . Thus, one can envision that core/shell NCs with a low-cost metal core and an ordered intermetallic SAA shell represent an innovative structural concept and hold promise for many catalytic applications. Nevertheless, it remains challenging to synthesize the precisely tailored nanoscopic architecture of such a well-defined structure.
To address this challenge, and as a proof-of-concept, here we report a facile, direct solution-phase synthesis of Cu/CuAu core/shell NCs with tunable, structurally ordered intermetallic SAA layers. The synthesis is also applicable to other Cu/CuM (M = Pt, Pd) systems. Using monodisperse copper nanocubes as templates and through a seed-mediated colloidal method, we overcome the kinetic barrier for atom diffusion by creating an additional driving force (the Kirkendall effect driven by galvanic replacement 27,28 ) other than thermal energy to promote ordered intermetallic formation, lowering the reaction temperature to make it compatible with organic solvents and ligands. We then synthetically controlled the density of single atoms and systematically investigated the ligand and strain effects of those Cu/CuAu NCs Article https://doi.org/10.1038/s44160-023-00258-x a d-spacing of ~0.36 nm in the particle core, corresponding to the (100) planes of the face-centred cubic (fcc) copper structure (ICSD #15985). The atomically dispersed gold atoms in the copper matrix on the shell can be directly visualized with a brighter contrast in the HAADF images, indicating the formation of a dilute SAA. The corresponding fast Fourier transform (FFT) pattern (Fig. 2d) of the nanocube in Fig. 2c confirmed that the Cu/CuAu dilute SAA has maintained the fcc copper structure. In contrast, Cu/CuAu ordered SAA nanocubes with ~7.5 wt% gold and similar particle size as the Cu/CuAu SAA (Fig. 2e,f) demonstrate the formation of the tetragonal P4/mmm intermetallic structure of the CuAu shell which can be directly observed from the HAADF images ( Fig. 2g and Supplementary Fig. 2) and the superlattice points (red circles) in the corresponding FFT pattern ( Fig. 2h and Supplementary  Fig. 3). The elemental maps of gold and copper in Fig. 2i show that gold atoms are homogeneously distributed on the surface of the nanocubes while copper atoms are distributed across the entire nanocubes, which is consistent with the particle morphology with an intermetallic CuAu shell and a copper core. We emphasize that in such an intermetallic CuAu structure, gold atoms are completely isolated by copper atoms (Supplementary Fig. 4).
It is also important to note that the CuAu layers form coherent interfaces with the copper core. Since the bulk CuAu intermetallic phase (ICSD #42574) has a tetragonal structure with lattice parameters (a = b ≈ 2.80 Å, c ≈ 3.67 Å) larger than that of the copper, the CuAu layer    Fig. 2k,l, respectively. The strains relative to the CuAu bulk lattice parameters are plotted in Fig. 2m. At the interface (atomic layer #4), the initial strains in both directions are around 8-10%, and the strain gradually decreases moving toward the surface of the particle. Although it may vary from place to place, the data shown in Fig. 2k-m suggest that around 1-3% compressive strain probably remains at the top surface of the CuAu layer. In contrast, in the case of dilute SAA on copper nanocubes, the copper lattice remains largely unchanged, except for some local variations due to the presence of bigger gold atoms ( Supplementary Fig. 5).
The powder X-ray diffraction pattern of the Cu/CuAu ordered SAA nanocubes demonstrates diffraction peaks that match well with the copper fcc phase ( Supplementary Fig. 6). The Panalytical X-ray diffractometer could not detect the signal of the ordered intermetallic CuAu shell due to this being only a few atomic layers thick. Therefore, synchrotron X-ray diffraction was performed to characterize the crystal structure of Cu/CuAu SAA nanocubes. As shown in Fig. 3a, in addition to those peaks detected in both Cu/CuAu SAA and copper nanocubes, an additional peak at 2θ = 15.6 was observed, which can be assigned to the (101) peak of P4/mmm ordered intermetallic CuAu structure, consistent with our STEM results. Energy-dispersive X-ray spectroscopy results ( Supplementary Fig. 7) show that the gold weight ratio is about 7.8 wt% for ordered Cu/CuAu SAA nanocubes, consistent with the results from inductively coupled plasma optical emission spectrometry.
The X-ray photoelectron spectroscopy (XPS) results in Supplementary Fig. 8 reveal that the copper and gold in Cu/CuAu ordered SAA nanocubes both correspond to metallic states. The electronic interactions between gold and copper were investigated via X-ray absorption near-edge spectroscopy (XANES). Figure 3b shows the normalized gold L 3 -edge XANES spectra of Cu/CuAu SAA, Cu/CuAu ordered SAA nanocubes and the reference (gold foil). In comparison with the gold foil, the Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes show a shift of absorption edge to the lower energy, indicating charge transfer from copper to gold. In the XANES post-edge region, the Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes show lower intensity peaks than the gold foil, which could be attributed to the single-site gold isolated by copper atoms. Figure 3c presents the Fourier-transformed gold L 3 -edge extended X-ray absorption fine structure (EXAFS) spectrum of the Cu/CuAu SAA, Cu/CuAu ordered SAA nanocubes and gold foil. In the R space, Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes exhibit a prominent peak at ~2.27 Å from the Au-Cu bonds 35 , while no typical peaks for Au-Au bonds at a higher R-value (~2.52 Å) appear 36 . EXAFS fitting was further conducted to obtain a quantitative structural configuration of gold (Supplementary Figs. 9-11 and Supplementary Table 1). The Cu/CuAu SAA (~2.64 Å) and Cu/CuAu ordered SAA nanocubes (~2.69 Å) exhibit shorter interatomic distance R Au-Cu(Au) than that of fcc gold atoms in gold foil (~2.86 Å), and the Au-Au coordination is absent in Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes. These results confirm that the gold atoms in Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes are present in single atomic dispersion. The coordination number (CN) of gold atoms in Cu/CuAu ordered SAA nanocubes is 7.28, which is smaller than that in the bulk gold foil (12), consistent with the theoretical CN of 8 for ordered CuAu intermetallic structure and 12 for the gold fcc structure. For the Cu K-edge, the Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes exhibit higher white-line intensity than the copper foil, which indicates the electron deficiency of the copper atoms in Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes due to the electron transfer from copper to gold (Supplementary Fig. 12a). There is no obvious shift in the R space due to the presence of bulk copper in the core region ( Supplementary Fig. 12b). EXAFS fitting results confirm that there are some Cu-Au bonds in Cu/ CuAu SAA and Cu/CuAu ordered SAA nanocubes (Supplementary Figs.  13-15 and Supplementary Table 2). The CN of copper atoms in Cu/CuAu ordered SAA nanocubes consists of CN Cu-Cu (7.25) and CN Cu-Au (2.74), which is due to the presence of both the bulk copper in the core and the ordered CuAu alloy in the shell. The wavelet transform of Au L 3 -edge EXAFS oscillations was performed and the corresponding contour plots (Fig. 3d-f) demonstrate intensity maxima at ~10.4 Å −1 of gold foil that can be attributed to the Au-Au contribution 36 . In contrast, the wavelet transform contour plots of Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes display maximum intensity at ~7.0 Å −1 and ~6.8 Å −1 , which should be associated with the Au-Cu bonding. Taken cumulatively, the above results indicate that the gold sites are atomically dispersed in Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes. The developed synthetic methods were further extended to precisely control the deposition of ordered intermetallic CuPd on copper nanocubes (Cu/CuPd ordered SAA) and ordered intermetallic Cu 3 Pt on copper nanocubes (Cu/Cu 3 Pt ordered SAA), demonstrating the generality of the synthesis. As shown in Supplementary Fig. 16a,b, the HAADF image demonstrates the formation of an ordered B2 CuPd intermetallic structure shell on a copper core, while the Cu/Cu 3 Pt shows an ordered L1 2 Cu 3 Pt intermetallic structure and platinum single atoms on the surface ( Supplementary  Fig. 16c,d). The difference in the ordered intermetallic structure formed in Cu/CuAu, Cu/CuPd and Cu/Cu 3 Pt is rooted in the thermodynamic stability of different phases, galvanic reaction kinetics and interdiffusion rates between copper and palladium, platinum and gold atoms.

NO 3 RR on core/shell Cu/CuAu SAA catalysts
We recently reported that the ordered intermetallic CuPd structure could break the adsorption-energy scaling limitations of the electrocatalytic NO 3 RR to NH 3 by engaging the interatomic coupling from the subsurface ligand 37 . We envision that such an effect can be inherited from the ordered intermetallic structure in our core/shell systems with largely improved precious metal atom efficiency. To evaluate the electrocatalytic NO 3 RR activities of the Cu/CuAu SAA and Cu/CuAu ordered SAA nanocubes, the catalysts were loaded onto carbon black (Vulcan XC-72R) ( Supplementary Fig. 17). As a comparison, copper nanocubes with a size of ~40 nm ( Supplementary Fig. 1) and gold nanocubes with a size of ~50 nm ( Supplementary Fig. 18) were synthesized. The polarization curves for the NO 3 RR obtained with Cu/CuAu SAA, Cu/CuAu ordered SAA nanocubes, copper nanocubes and gold nanocubes are shown in Fig. 4a. The Cu/CuAu ordered SAA nanocubes show an onset potential of 0.26 V versus RHE, much more positive than that of Cu/ CuAu SAA nanocubes (0.20 V), copper nanocubes (−0.10 V) and gold nanocubes (−0.18 V). The partial current density of NH 3 on Cu/CuAu ordered SAA nanocubes is also higher than that on Cu/CuAu SAA, copper and gold nanocubes (Fig. 4b). For the hydrogen evolution reaction (HER), Cu/CuAu SAA and Cu/CuAu ordered SAA show activities slightly higher than that of copper but lower than that of gold ( Supplementary  Fig. 19). The double-layer capacitance (C dl ) of the Cu/CuAu ordered SAA nanocubes is calculated to be 4.75 mF cm −2 , which is close to those of Cu/CuAu SAA nanocubes (4.91 mF cm −2 ) and copper nanocubes (4.95 mF cm −2 ) due to the similar size of the catalysts ( Supplementary  Figs. 20 and 21). The ECSA-normalized current densities and partial current densities of NH 3 on Cu/CuAu ordered SAA nanocubes are much higher than that on Cu/CuAu SAA nanocubes, copper nanocubes and gold nanocubes, demonstrating that the intrinsic activity for the NO 3 RR toward NH 3 on Cu/CuAu ordered SAA nanocubes is superior to those on Cu/CuAu SAA, copper and gold nanocubes ( Supplementary Fig. 22).
Chronoamperometry measurements of catalysts were conducted at different potentials for 1 h in 1 M KOH + 1 M KNO 3 solution Article https://doi.org/10.1038/s44160-023-00258-x ( Supplementary Fig. 23). Gas chromatography was employed to detect the quantity of the gas product and only very little H 2 was identified from the competing HER. Ion chromatography was used to quantify produced NO 2 − (Supplementary Fig. 24), and the colorimetric method using Nessler's reagent (Supplementary Fig. 25) was used to quantify produced NH 3 . Faradaic efficiency and NH 3 yield rates of the catalysts are shown in Fig. 4c Table 3) [38][39][40][41] . The excellent performance of the NO 3 RR to NH 3 with >500 mA cm −2 is slightly inferior to one NO 3 RR catalyst, highly dispersed ruthenium atoms on copper nanowires, which possesses extremely high specific surface area and was tested in a flow-system hydrogen cell 29 . The main byproduct of the NO 3 RR on Cu/CuAu ordered SAA nanocubes is NO 2 − , as detected and quantified by ion chromatography (Supplementary Fig. 26). Control experiments were performed at −0.6 V versus RHE in 1 M KOH solution without KNO 3 . As shown in Supplementary Fig. 27, almost no NH 3 was detected in the electrolyte. 15 N isotope labelling experiments were carried out to further confirm that the produced NH 3 was derived from the feeding nitrate electrolyte. After electrolysis at −0.6 V versus RHE, no triple coupling peaks representing 14 NH 4 + were detected in the 1 H NMR spectra of the electrolyte, whereas doublet peaks of 15 NH 4 + were observed (Fig. 4e), confirming that the produced NH 3 originated from the electroreduction of nitrate. As a comparison, CuAu alloy nanoparticles with a size of ~10 nm were synthesized (Supplementary Fig. 28). As shown in Supplementary Fig. 29, CuAu alloy nanoparticles demonstrate a partial NH 3 current density, Faradaic efficiency of NH 3 and NH 3 yield rate lower than those of Cu/CuAu ordered SAA and Cu/CuAu Article https://doi.org/10.1038/s44160-023-00258-x SAA nanocubes, which indicates that the enhanced electrocatalytic performance of NO 3 RR to NH 3 originates from the controlled hosted single-atom sites in the shell.
To investigate the origin of the high performance of Cu/CuAu ordered SAA nanocubes, online DEMS was carried out to identify the intermediates and products (Fig. 4f). The m/z signals of 30, 17, 16, 15 and 14 that correspond to NO, NH 3 and fragments of NH 3 appeared during three continuous cycles from 0.4 V to −0.8 V versus RHE. Since NO was detected in the product gas stream, it is considered an important intermediate in the NO 3 RR pathway on our catalysts. Moreover, in situ attenuated total reflectance surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS) was also performed to detect the intermediates on Cu/CuAu ordered SAA during the NO 3 RR process. The absorption at around 1,645 cm −1 can be assigned to the H-N-H bending of NH 3 or the H-O-H bending of water molecules 42 . The absorption at around 1,370 cm −1 can be attributed to the adsorption of NO 3 − (ref. 43 ) (Supplementary Fig. 30). The absorption at around 1,109 cm −1 is due to the adsorption of NH 3 (ref. 42 ). The N=N stretching band at ~2,010 cm −1 did not appear, indicating that N 2 H x is not a likely intermediate of the NO 3 RR on Cu/CuAu ordered SAA 44 . The NO 3 RR performance on Cu/ CuAu ordered SAA nanocubes in a less concentrated KNO 3 solution demonstrated a slightly higher Faradaic efficiency toward NH 3 and lower current density and yield than that in 1 M KNO 3 ( Supplementary  Fig. 31). The stability of catalytic NO 3 RR performance was evaluated on Cu/CuAu ordered SAA nanocubes by conducting 20 consecutive electrolysis cycles at a fixed potential (Fig. 4g). The Faradaic efficiency of NH 3 exhibited a negligible decrease in 20 consecutive electrolysis cycles. As shown in Supplementary Fig. 32, after 12 h of continuous operation, there is no apparent decrease in the current density. After the stability test, the catalysts were removed from carbon paper by sonication and characterized by HAADF-STEM and XPS. As shown in Supplementary Fig. 33, for most of the particles, the cubic morphology was maintained after the long-term stability test. However, we did observe that in some particles, the copper core partially leached out during this process. This can be attributed to the nature of copper's tendency to dissolve/restructure during long-term electrochemical 6 45 and there might exist some defect sites on the few-atomic layers of ordered SAA, which could serve as the site for leaching out copper from the core. Nevertheless, the ordered CuAu layers remained in both cases, indicating the structural stability of the ordered SAA on copper. Moreover, XPS analysis demonstrates that there is no apparent shift in the binding energy of both gold 4f and copper 2p core levels before and after the stability test, suggesting that no chemical states changed after the stability test ( Supplementary Fig. 34).

Computational insights into the NO 3 RR on Cu/CuAu SAA nanocubes
DFT calculations were performed to gain insights into the high performance of the NO 3 RR of the Cu/CuAu ordered and dilute SAA nanocubes. The Cu/CuAu ordered SAA was modelled by cutting the copper-terminated (100) slab from the body-centred tetragonal (bct) P4/mmm CuAu, while for the Cu/CuAu SAA we took the Cu(100) surface and replaced a surface copper atom with a gold atom, as shown in Fig. 5a. Copper termination is used because this results in lower surface energy than gold termination with surface species, for example, *NO 3 . Further details on how each system was modelled and additional computational details are shown in Methods. While various reaction pathways have been suggested previously, we used the reaction pathway from our previous work which is the most thermodynamically favourable pathway on Cu(100) (ref. 37 ). Figure 5b shows an activity map at 0 V versus RHE using *NO 3 and *N binding energies as two reactivity descriptors. The activity is measured based on the maximum free energy barrier along the reaction pathway. The aforementioned surfaces are marked on this activity map (gold is far beyond the energy range and thus not shown) and the Cu/CuAu ordered SAA nanocubes show the highest activity, attributed to a stronger *NO 3 binding and weaker *N binding than Cu(100). Further analysis of the electronic structures shows that the d-band centre of copper atoms for the bct CuAu surface is higher than that of other surfaces, therefore leading to stronger binding of *NO 3 . The upshift in the d-centre also increases the hybridization contribution to the *N binding interaction as predicted by the traditional d-band theory. However, the subsurface gold interacts in a primarily repulsive manner because of a fully occupied d-band and large d-orbital radius, and thus the overall interaction is weaker, leading to a facile removal of nitrogen-bonded species. Fig. 2k,i show that the CuAu intermetallic surface has a compressive strain (1-3%). To evaluate the effect of the strain on the activity, the CuAu intermetallic with a 2.5% compressive strain has been plotted to show the predicted activity. It can be seen that the strain slightly strengthens the binding of *N and weakens the binding of *NO 3 , resulting in a negligible effect on the activity. Nevertheless, the strained system is predicted to be more active than the dilute SAA and the Cu(100) surfaces, indicating the dominant role of the ligand effect. The characterization of the Cu/ CuAu ordered SAA nanocubes suggested that there also exist (110) surfaces that are unlikely to be active since they bind *NO 3 too weakly (~ −1.50 eV) (Supplementary Table 4).

Conclusions
We designed and synthesized a catalytic motif, a core/shell dilute SAA and an ordered SAA, directly in the solution phase without solid-state thermal treatment. The d-state hybridization in the ordered CuAu alloy layers strengthened the *NO 3 binding, and the Pauli repulsion from the fully occupied metal d-states led to a weaker *N binding. The surface strain on Cu/CuAu ordered SAA exhibited a negligible effect on tuning the adsorption of key NO 3 RR intermediates. As a result, the exceptional ammonia selectivity and yield from the NO 3 RR on these Cu/CuAu ordered SAA catalysts represent remarkable progress in the development of SAA, ordered intermetallic and core/shell catalysts. This study opens an avenue for fabricating tunable, high-loading SAA catalysts by a direct solution synthesis for sustainable electrocatalytic reactions.  Article https://doi.org/10.1038/s44160-023-00258-x reagent and Nafion (5 wt%) were all purchased from Sigma-Aldrich. Hexane and ethanol were technical grade from Sigma-Aldrich and were used without further purification.

Synthesis of Cu/CuAu SAA and ordered SAA nanocubes
In a typical procedure, 0.2 mmol CuBr, 0.5 mmol TOPO and 10 ml OAm were added into a 50 ml three-necked flask under stirring. The mixture was heated under N 2 atmosphere to 80 °C and kept at this temperature for 30 min. Then the mixture solution was heated to 260 °C at a heating rate of 10 °C min −1 and incubated at this temperature for 30 min, generating a reddish solution. After cooling down to 200 °C, 0.5 ml or 1 ml of 0.01 mmol ml −1 HAuCl 4 OAm solution was dropwise injected into the solution and incubated at this temperature for 30 min. After cooling down to room temperature, the precipitate was centrifuged and washed three times with hexane and excess ethanol and dispersed in hexane. Cu/CuPd ordered SAA and Cu/Cu 3 Pt ordered SAA were synthesized by similar procedures to the synthesis of Cu/CuAu ordered SAA except that the precursors were changed to Pd(acac) 2 and Pt(acac) 2 .

Synthesis of copper nanocubes
In a modified procedure 33 , 0.2 mmol of CuBr, 0.5 mmol TOPO and 10 ml OAm were loaded into a 25 ml three-necked flask under stirring. The mixture was heated under N 2 atmosphere to 80 °C and kept at this temperature for 30 min. The mixture solution was then heated to 260 °C at a heating rate of 10 °C min −1 and incubated at this temperature for 30 min, generating a reddish solution. After cooling to room temperature, the precipitate was centrifuged and washed three times with excess ethanol and dispersed in hexane.

Synthesis of gold nanocubes
In a modified procedure 46 , 0.25 ml of 10 mM HAuCl 4 ·3H 2 O aqueous solution was added into 7.5 ml of 100 mM CTAB aqueous solution, which was gently mixed by magnetic stirring. Subsequently, 0.60 ml of 10 mM of ice-cold freshly prepared NaBH 4 aqueous solution was added, followed by magnetic stirring for 2 min. The solution was then left undisturbed at 25 °C for 1 h to obtain gold seeds. Then, 2.0 ml of 10 mM HAuCl 4 ·3H 2 O aqueous solution was added to the mixture containing 16 ml of 100 mM CTAB solution and 80 ml of deionized water. Next, 9.5 ml of 100 mM l-ascorbic acid aqueous solution was added. Finally, 50 μl of the 1 h aged, diluted gold seed solution (diluted 10 times with deionized water) was added to the solution, which was kept undisturbed for 1 h. The product was collected by centrifugation, washed several times with deionized water and dispersed in deionized water.

Synthesis of CuAu alloy nanoparticles
In a typical procedure, 0.1 mmol Cu(acac) 2 , 0.1 mmol HAuCl 4 ·3H 2 O and 10 ml OAm were loaded into a 25 ml three-necked flask under stirring. The mixture was heated under N 2 atmosphere to 80 °C and kept at this temperature for 30 min. The mixture solution was then heated to 200 °C at a heating rate of 10 °C min −1 and incubated at this temperature for 30 min. After cooling to room temperature, the precipitate was centrifuged and washed three times with excess ethanol and dispersed in hexane.

Preparation of carbon-supported catalysts (20% loading)
To prepare carbon-supported catalysts, we mixed a hexane dispersion of 20 mg of catalysts with 80 mg activated carbon (Vulcan XC-72R) and sonicated for 2 h. The catalysts were collected by centrifugation, washed three times with excess ethanol and dried for 8 h in a vacuum oven at 60 °C.

Characterization
X-ray diffraction was performed on a Philips X'Pert PRO SUPER with Cu Kα (λ = 1.54056 Å). XPS was performed on a PHI Versa Probe III scanning XPS microscope using a monochromatic Al Kα X-ray source (1,486.6 eV). The morphology was characterized by TEM (EM-420). BF-TEM, HAADF-STEM and energy-dispersive X-ray spectroscopy were conducted on a JEOL ARM 200CF equipped with an Oxford Instruments X-ray energy-dispersive spectrometer. The elemental contents of the Cu/CuAu SAAs were determined by inductively coupled plasma optical emission spectrometry on a SPECTRO GENESIS inductively coupled plasma spectrometer. The colorimetric method with Nessler's reagent on an ultraviolet-visible spectrophotometer (Agilent 3500) was used to quantify the produced ammonia. The gas product was quantified by gas chromatography (Agilent 7890B). Ion chromatography (Metrohm Eco IC) was used to quantify the produced nitrite. Liquid products were analysed by 1 H NMR using dimethyl sulfoxide as an internal standard. The X-ray absorption spectra of gold and copper K-edges were obtained at the beamline 12−BM-B station of the Advanced Photon Source at Argonne National Laboratory. Both gold L 3 -edge and copper K-edge XANES and EXAFS were measured under fluorescence mode by a Vortex ME4 detector. All XAS data analyses were performed with the Athena software package to extract XANES and EXAFS. Synchrotron X-ray diffractiondata were collected at 20 keV (λ = 0.6199 Å) at beamline 7-BM of NSLS-II, Brookhaven National Laboratory. The distance from detector to the sample was 299.48 mm, calibrated with LaB6. The twodimensional X-ray diffraction patterns were integrated with Dioptas software.

Electrochemical measurements
Electrochemical measurements were performed using a three-electrode system connected to a BioLogic electrochemical workstation. All measurements were performed at room temperature in a gas-tight hydrogen cell separated by an ion-exchange membrane (Nafion 117 The concentrations of NO 2 − were determined by ion chromatography (Metrohm Eco IC).

DEMS measurements
DEMS was performed in a Type B cell (Hiden Analytical), using an Autolab Potentiostat (PGSTAT204, Metrohm). The electrocatalysts were drop-cast onto a titanium plate as the working electrode. Platinum wire and Ag/AgCl were used as the counter electrode and reference electrode, respectively. Electrolyte (1 M KOH + 1 M KNO 3 ) was kept flowing into the electrochemical cell with a flow rate of 0.5 ml min −1 through a peristaltic pump. Argon was bubbled into the electrolyte constantly before and during the DEMS measurements. LSV technology was employed from −0.6 to −1.8 V versus Ag/AgCl with a sweep rate of 20 mV s −1 . Mass spectra were acquired on a Hiden HPR40 (Hiden Analytical) dissolved-species mass spectrometer. An electron energy of 70 eV was used for ionization of all species, with an emission current of 450 μA. All mass-selected products were detected by a secondary electron multiplier with a detector voltage of 1,200 V. After the mass signal returned to baseline at the end of the electrochemical test, the next cycle was started using the same test conditions. The experiment ended after three cycles.

In situ ATR-SEIRAS experiments
ATR-SEIRAS experiments were conducted in a one-compartment PEEK spectroelectrochemical cell with the VeeMAXIII ATR accessory. A graphite rod counter electrode and a saturated Ag/AgCl reference electrode controlled by a Autolab Potentiostat (PGSTAT204, Metrohm) were used. Infrared measurements were performed in a Thermo Nicolet iS50 FTIR equipped with a liquid-nitrogen-cooled MCT detector. A modified chemical deposition method was used to generate gold film electrodes on the reflecting plane of silicon ATR crystal prisms cut to 60° incidence. The experiments were performed in 0.1 M KOH + 1 M KNO 3 . Thirty-two scans with a resolution of 4 were averaged at each potential with a sampling frequency of 20 s at 5 mV s −1 .

DFT calculations
DFT calculations were performed using the Vienna Ab initio Software package 47,48 with projector augmented wave psuedopotentials. The exchange-correlation interaction was treated at the generalized gradient approximation level using the revised Perdew-Burke-Ernzerhof functional 49 . A planewave energy cutoff of 450 eV was used and the Methfessel-Paxton smearing scheme was used with a smearing parameter of 0.2 eV. Electronic energies are extrapolated to k B T = 0 eV. Because of the challenges with calculating electrochemical barriers and the universality of the Brønsted-Evans-Polanyi principle, we used thermodynamic barriers to rationalize the activity trend. The CuAu ordered SAA was modelled by generating (100) and (110) slabs from a P4/mmm ordered intermetallic CuAu bulk. The lattice constants for this bulk were optimized via DFT to be a = 2.86 Å and c = 3.66 Å. A 3 × 3 × 6 supercell with the bottom four layers fixed was used for the (100) slab and a 3 × 3 × 4 supercell with the bottom two layers fixed was used for the (110) slab. The Brillouin zones for the (100) and (110) slabs were sampled using Monkhorst-Pack meshes of 4 × 4 × 1 and 2 × 3 × 1, respectively. An additional model system was generated to consider a compressive strain shown in Fig. 2k,i. The CuAu bulk was reoptimized but with the one of lattice constants compressed by 2.5% (a = 2.79 Å), and a compressed (100) slab was generated from this bulk. The CuAu SAA was modelled by replacing a single surface copper atom in a Cu(100) slab with a gold atom. The slab was a 4 × 4 × 4 supercell with the bottom two layers fixed and the Brillouin zone was sampled using a 4 × 4 × 1 Monkhorst-Pack mesh. To consider the effect of the gold atom, all adsorbates were assumed to bind at a site near the gold atom. The energetics for the copper nanocubes are the same as the values used for Cu(100) from our previous publication 37 . Grand canonical DFT calculations were performed to consider the solvation and electric field effects at the electrode/electrolyte interface. Adsorbate free formation energies were calculated similarly to our previous publication using the same corrections and references 37 .

Data availability
The data supporting the finding of the study are available in the paper and its Supplementary Information. Source data are provided with this paper.